This pull request changes some code to align with the decision "Singleton type RecentEra :: AP".

In some places, I have

• Removed an explicit RecentEra era argument in favor of the type class / implicit argument Write.IsRecentEra era — unless this would have resulted in an ambiguous type, such as for mkTransaction.
• Removed @era applications that are not of the form recentEra @era.
  (I have left applications of the form shelleyBasedEra @era and cardanoEra @era for convenience, as they are similar in spirit.)

Comments

• This pull request is an alternative to [Experimental] Simplify the use of recent eras #4409 .

• I have kept or even added an explicit RecentEra era argument to the functions buildTransaction, constructTransaction, and constructUnbalancedSharedTransaction, because their usage site (Server module) is prone to ambiguity. The era type is introduced through an existential like this

          (Write.PParamsInAnyRecentEra era pp, _)
              <- liftIO $ W.readNodeTipStateForTxWrite netLayer

  In order to get the era type into scope, we would need to write

          (Write.PParamsInAnyRecentEra (_era :: RecentEra era) pp, _)
              <- liftIO $ W.readNodeTipStateForTxWrite netLayer

  In both cases, the value or type are needed to specify in which era the SealedTx should be created.

Issue Number

ADP-3226